[R] Access variables by string

[Peter Dalgaard]

Philipp Rappold wrote:
> Dear all,
> 
> I have two probably very easy questions:
> 
> (1) Is there a way to access certain variables by their string-based
> name representation?
> 
> Example:
> numbers <- c("one", "two", "three")
> varname <- "numbers"
> print(varname[2])
> 
> (2) I need this functionality for a customized na.exclude() function
> that I am building, which should only exclude rows that have NA in
> certain columns. Maybe there is already a function which does exactly
> what I need, so I'd highly appreciate if someone could point me there ;)
> 
> My current implementation looks like this:
> 
> naexlcude <- function(data, varnames)
> {
>     for(v in varnames){
>         data = subset(data, !is.na(v))
>     }
> 
>     data
> }
> 

Well, you can use get(varname)[2] or more generally things like
eval(bquote(.(as.name(varname))[2])), but for this particular
application, why not just use the standard indexing techniques?

I think this will do

ix <- apply(is.na(data[varnames]), 1, any)
data[!ix,]

e.g.

> table(apply(is.na(airquality[c("Wind","Ozone")]),1, any))

FALSE  TRUE
  116    37
> colSums(is.na(airquality))
  Ozone Solar.R    Wind    Temp   Month     Day
     37       7       0       0       0       0
> table(apply(is.na(airquality[c("Solar.R","Ozone")]),1, any))

FALSE  TRUE
  111    42


(Using functions like subset inside another function often leads to
problems because of the nonstandard evaluation tricks that it uses. It
is mainly useful to save tying on the command line.)

-- 
   O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark      Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)              FAX: (+45) 35327907